About

Project Overview

This project explores the predictive modeling of short-term volatility anomalies in the Chinese equity market, with a specific focus on the stock AtHub (603881.SH)—a data center infrastructure company. The goal is to develop a machine learning pipeline that can detect abnormal daily price or volume movements using technical analysis (TA) indicators.

We define anomalies as:

  • Price spikes or crashes: Daily returns exceeding $$5%.

  • Volume surges: Daily trading volume exceeding 2× the 30-day moving average.

By engineering over 30 TA-based features (momentum, trend, volume, and volatility), we aim to identify leading patterns that consistently precede such events.

Approach Summary

  • Data Collection: Daily OHLCV data for AtHub over a multi-year period.
  • Feature Engineering: Over 30 technical indicators computed using rolling windows.
  • Modeling: Supervised learning for anomaly classification (binary targets).
  • Evaluation: Time-aware cross-validation to preserve chronological integrity.
  • Interpretation: SHAP analysis to understand key feature interactions.

Contributors

This project was developed by

  • Annabelle Zhu, Project Author
  • Dr. Greg Chism, Professor of INFO 523.